Blog

The Digital Space of Berlin’s Housing Market: A Look at Twitter 

12. July 2024

“Nice and loud protest outside of #Meuterei in #Kreuzberg! Great weather and good atmosphere! #MeutereiStays #SyndikatStays #PotseStays #DrugstoreStays #Liebig34Stays #g17Stays #RummelbuchtStays #DieselA #WeAreAllStaying #RentMadness” 

This message1 was posted to Twitter by a Berlin-based housing activist, presumably while on the ground protesting the impending eviction of a leftist bar, Meuterei. In their message, through the use of a list of hashtags, the activist not only explicitly tied the struggle to the district of Kreuzberg, they also connected it to several other places of resistance against gentrification in the city. Through naming places and tying them to issues and actions, the Twitter user constructed a discursive space. 

On a different day, and in a starkly different tone from the activist’s enthusiasm in the first example, another user tweeted: 

“The problem with Spandau (where I work) is that over the past few years, more and more ‘problematic tenants’ from the lower classes moved there. Also an effect of the explosion of rents in the city center. We are creating a Berlin Banlieau, only no one is noticing it yet.” 

Gentrification, this user claimed, was creating quarters with a high concentration of poverty in the city’s outskirts. This pattern would lead to social frictions, similar to those popularly associated with Paris. 

Each of these Twitter messages reflects only a snapshot of one user’s perspective at a single point in time. Whether one agrees or disagrees with their assessments of how the issue of housing plays out across Berlin’s urban space, one single Twitter message will have limited effects and disappear into digital nirvana in short order.  

However, they represent two messages out of tens of thousands. A large number of users routinely take to social media to express their ideas about the issue. Users range from dedicated, strategic activists posting frequently, to members of the public sharing experiences, and to political actors informing on measures in local government. Such an ensemble of actors coming together to shape “public imagination”2 of an issue is commonly termed an “issue public”3. Crucially, these actors not only share ideas of the ‘what’ and ‘why’ of public issues, they also tie them to the question of ‘where’. This dimension of the public construction of issues has not received much attention. 

In my recent book, I was interested in the patterns and ideas about the spatiality of issues, which emerged in social media discourses. Moreover, I wanted to understand why different places received the levels and forms of public attention they did. Could the prevalence of activist communication, contentious political debates, or stigmatized narratives on social media be connected back to local resources and infrastructures, such as the density of local organizations or the residential composition?  

To answer these questions, I brought together scholarship on public spheres, social inequalities, and spatial sociology to develop the concept of ‘issue spaces’. An issue space emerges over time because public actors practice “place-naming”4 to tie together places and social problems. If these instances of place-naming are not random, but follow particular, unequal patterns, they will shape ideas about where the issue is or is not prevalent, where it is being worsened, or where it is being addressed. 

To put this idea into practice, I started out with a large corpus of Twitter messages, which referred to housing and the city of Berlin in some way. In the first step, I automatically applied a gazetteer (a dictionary of geographical entities) to identify references to Berlin’s 96 districts, around 12,000 streets, and a number of more informal neighborhoods (“Kieze”) in the tweets’ texts. This narrowed the data down to around 30,000 tweets from 2018 and 2019, each of which discussed at least one specific location in the city. 

As a first step of analysis, I then mapped the distribution of these references across the city (Figure 1). Even after adjusting for population size to ensure that patterns are not an effect of larger districts being discussed more frequently, a stark picture emerges. Public attention on social media reveals a large gap between center and periphery. Places in the urban core are the subject of thousands of messages, whereas most places in the periphery are only discussed a handful of times over the two-year investigation period. The most ‘visible’ district, seen with the darkest red – Kreuzberg – was discussed, per capita, about 160 times more frequently than the least visible places. 

Figure 1. The distribution of place-naming in Berlin. (Source: Daniela Stoltenberg)

Observing references to districts, streets, and informal neighborhoods separately, although a center-periphery gap emerges in each map, it is not uniform. High attention (visibility) is more dispersed when it comes to discussions at the district level than at the street level. In other words, when more peripheral places are discussed on Twitter, it is likely within the broad framework of whole districts. Discussions of events at the street level – a protest is held, a building is evicted – are more concentrated to the center.  

Focusing on informal neighborhoods, what jumps out the most is an example of how concerted social media action can disrupt typical, macro patterns. Above all, Altglienicke in the southeastern part of the city stands out. This is due to posts discussing the Kosmosviertel neighborhood. Here, residents mobilized against luxury renovations to the buildings, which were ultimately remunicipalized (reversely privatized) by the city government. While on a macro scale, social media attention is unequally distributed to privilege central locations, we also see evidence of its opportunity structures to mobilize activists in the periphery. 

Regression models including different socio-spatial indicators to explain patterns of attention allocation reveal a complex picture. Strikingly, even though we see a clear center-periphery gap, location centrality is not in itself predictive of attention. Rather, places with high visibility tend to have lower proportions of marginalized populations, especially elderly and poor residents. Places with high visibility tend to have a high density of community organizations and meeting places, which may be leveraged for organizing and “neighborhood storytelling”5. This means that a typical place receiving a lot of attention concerning the housing crisis has relatively young and economically well-off residents, and a large number of local organizations in which they can meet and participate. These places are not necessarily where housing prices have risen particularly quickly. In essence, who is able to mobilize seems to have a larger effect on visibility than how big the problem is, locally. 

Although differences in the volume of public attention are already striking, how people imagine the issue space will not only be influenced by which places receive attention but also how. To map these differences in the way places around Berlin are discussed in the context of housing, I calculated a so-called “structural topic model”6. Topic modeling is an inductive, computational approach which detects patterns in word occurrences in large text corpora. Based on lists of important keywords and documents, researchers can interpret the results of this procedure as topics or sub-issues. I combined the results of topic modeling with the place-naming data discussed in the previous step.  

In total, ten sub-issues were found in the Berlin housing Twitter data. Figure 2 shows three examples, which illustrate overall patterns well.7 As shown in the map of Topic 2, which consists of tweets about anti-gentrification protests, sub-issues around activism are strongly clustered in the urban center. This is consistent within topics ranging from squatting activism, to larger protest campaigns, and local struggles around particular buildings. Activism is very much a seen as a phenomenon of the city center in the digital sphere. 

Figure 2. The distribution of sub-issues: Three example topics. (Source: Daniela Stoltenberg)

To a lesser extent, this pattern is also found in most topics associated with politics and policy. These sub-issues, such as Topic 6 about local tenant protection measures, contain controversial debates about how to best address Berlin’s housing crisis. Often, there is back and forth between local political actors and residents or small organizations.  

One major exception is Topic 5. Tweets related to this sub-issue discuss affordable housing policy, and especially new construction as a way to create affordable housing. It is one of very few topics which are more strongly associated with places in the periphery (and Tempelhofer Feld8). Here, debate is often driven by citywide political and economic actors. Unlike the previous topics, local voices from the places under discussion are scarce. The periphery is mostly discussed as an easy solution to the housing crisis by actors not located there. 

Overall, the investigation of sub-issues adds depth and quality to the observed differences in quantity. Central places are constructed as the locus of activism and contentious political debate. They are discussed with more specificity and more agency than those in the urban periphery. In a nutshell, housing is constructed as an issue that arises and is negotiated in the center, but can be solved in the periphery. The “issue space” is characterized by this dynamic. 

In the larger research project, I did not stick to only the question of housing in Berlin. I also investigated housing in another city, Frankfurt am Main, and a different issue – bicycle infrastructure – in both cities. Although differences arose in detail, many patterns can be observed for all of these issue spaces. Urban issue spaces on Twitter tend to be driven by inequality in attention, especially between center and periphery, and to some extent by a local clustering of attention. Some places receive outsized attention, which is driven by dedicated digital activists, and demonstrates that social media has a potential to draw attention to local struggles. Still, beyond these individual cases of hyperlocal visibility, issue spaces tend to highlight those places where relatively privileged populations live and where there is a great deal of infrastructure for organizing and civic mobilization. Twitter, by and large, highlights privileged places, it is not the place where socio-spatial marginality is challenged. 

Author Biography: Dr. Daniela Stoltenberg is a postdoctoral researcher at Freie Universität Berlin and the CRC 1265 in subproject B05 “Translocal Networks.” Her research interests lie in communication geography, digital public spheres, new social movements, and computational research methods. 

Further reading: Stoltenberg, D. (2024). The spaces of public issues: How social media discourses shape public imaginations of issue spatiality. Routledge. https://doi.org/10.4324/9781003428961


1 Messages originally in German, translated and lightly altered for anonymity. 

2 Asen, R. (2002). Imagining the public sphere. Philosophy and Rhetoric, 35(4), 345- 367. https://doi.org/10.1353/par.2003.0006

3 Bennett, W. L., Lang, S., & Segerberg, A. (2015). European issue publics online: The cases of climate change and fair trade. In T. Risse (Ed.), European public spheres: Politics is back (pp. 108-138). Cambridge University Press.

4 Wiard, V., & Pereira, F. H. (2019). Bad neighborhoods in a good city? Space, place and Brussels’ online news. Journalism Studies20(5), 649-674. https://doi.org/10.1080/1461670X.2017.1417052  

5 Ball-Rokeach, S. J., Kim, Y.-C., & Matei, S. (2001). Storytelling neighborhood: Paths to belonging in diverse urban environments. Communication Research, 28(4), 392-428.  https://doi.org/10.1177/009365001028004003 

6 Roberts, M. E., Stewart, B. M., & Tingley, D. (2019). Stm: An R package for structural topic models. Journal of Statistical Software91, 1-40. https://doi.org/10.18637/jss.v091.i02

7 Topics are especially prevalent in places marked red and absent in places marked blue. The darker the shade of color, the stronger this deviance is.

8 Tempelhofer Feld is a former airfield in the urban center, which is now used as a large public park. There have been heated debates, including a public referendum, about the possibility of building housing on parts of it.